Possibilities of artificial intelligence with the use of a structured algorithm in the diagnosis of keratoconus
DOI:
https://doi.org/10.31288/oftalmolzh202561317Keywords:
artificial intelligence, ChatGPT, keratoconus, corneal topographyAbstract
Purpose: To assess the efficacy of using artificial intelligence (AI) in the diagnosis of keratoconus based on corneal topography maps, with subsequent determination of disease stage and examining the impact of the level of standardization of the clinical algorithm on the accuracy and consistency of results.
Material and Methods: Fifty-nine corneal topography maps of 37 patients from the database of the ACE DIAGNOSTIC PLATFORM (Baush+Lomb) were retrospectively analyzed. The study was conducted in three stages which simulated various levels of AI integration, from a completely autonomous AI operation to the operation strictly according to a three-level structured algorithm that we have developed (Copyright for the Work No. 139012 issued 26.08.2025). The adapted language model of ChatGPT (OpenAI, 2023) was used for establishing the diagnosis and staging. Data were statistically analyzed using Microsoft Excel 2019 and Analysis ToolPak. Data analysis was conducted by solving binary and multi-class classification problems. Accuracy, sensitivity, specificity, precision and the F1-measure were assessed at each stage and compared across stages.
Results: The overall accuracy and accuracy of staging were 79.7% and 33.3%, respectively, at stage 1, and improved to 93.2% and 74.36%, respectively, at stage 2. The highest values of characteristics were seen at stage 3 (sensitivity, 100%; specificity, 95%; overall accuracy, 98%; and accuracy of staging, 89.7%).
Conclusion: AI demonstrated a potential in the automated diagnosis of keratoconus. However, the best efficacy was achieved only when AI was used in the presence of a clearly structured algorithm. Autonomous AI decision led to a high variation in results. Introduction of algorithmic logic enabled significant improvements in the accuracy and consistency of results.
References
Santodomingo-Rubido J, Carracedo G, Suzaki A, Villa-Collar C, Vincent SJ, Wolffsohn JS. Cont Lens Anterior Eye. 2022 Jun;45(3):101559.
https://doi.org/10.1016/j.clae.2021.101559
Lin SR, Ladas JG, Bahadur GG, Al-Hashimi S, Pineda R. A Review of Machine Learning Techniques for Keratoconus Detection and Refractive Surgery Screening. Semin Ophthalmol. 2019;34(4):317-326. https://doi.org/10.1080/08820538.2019.1620812
Ferdi AC, Kandel H, Nguyen V, Tan J, Arnalich-Montiel F, Abbondanza M, Watson SL. Five-year corneal cross-linking outcomes: A Save Sight Keratoconus Registry Study. Clin Exp Ophthalmol. 2023 Jan;51(1):9-18. https://doi.org/10.1111/ceo.14177
Bui AD, Truong A, Pasricha ND, Indaram M. Keratoconus Diagnosis and Treatment: Recent Advances and Future Directions. Clin Ophthalmol. 2023;16(17):2705-2718. https://doi.org/10.2147/OPTH.S392665
Niazi S, Gatzioufas Z, Doroodgar F, Findl O, Baradaran-Rafii A, Liechty J, Moshirfar M. Keratoconus: exploring fundamentals and future perspectives - a comprehensive systematic review. Ther Adv Ophthalmol. 2024 Mar 20:16:25158414241232258.https://doi.org/10.1177/25158414241232258
Li Z, Wang L, Wu X, Jiang J, Qiang W, Xie H, et al. Artificial intelligence in ophthalmology: The path to the real-world clinic. Cell Rep Med. 2023;4(7):101095. https://doi.org/10.1016/j.xcrm.2023.101095
Afifah A, Syafira F, Afladhanti PM, Dharmawidiarini D. Artificial intelligence as diagnostic modality for keratoconus: A systematic review and meta-analysis. J Taibah Univ Med Sci. 2023;19(2): 296-303.https://doi.org/10.1016/j.jtumed.2023.12.007
Wan Q, Wei R, Ma K, Yin H, Deng YP, Tang J. Deep Learning-Based Automatic Diagnosis of Keratoconus with Corneal Endothelium Image. Ophthalmol Ther. 2023 Dec;12(6):3047-3065. https://doi.org/10.3390/diagnostics
Muhsin ZJ, Qahwaji R, Al Shawabkeh M, et al. Smart decision support system for keratoconus severity staging using corneal curvature and thinnest pachymetry indices. Eye Vis (Lond). 2024;11(1):28. https://doi.org/10.1186/s40662-024-00394-1
Yousefi S, Yousefi E, Takahashi H, Hayashi T, Tampo H, Inoda S, et al. Keratoconus severity identification using unsupervised machine learning. PLoS One. 2018;13(11) e 0205998. https://doi.org/10.1371/journal.pone.0205998
Zhamardiy VO, Shkola OM, Okhrimenko IM, Strelchenko OG, Aloshyna AI, Opanasiuk FH, et al. Checking of the methodical system efficiency of fitness technologies application in students' physical education. Wiad Lek. 2020;73(2):332-341. https://doi.org/10.36740/WLek202002125
Nevska AO, Pohosian OA, Goncharuk KO, Chernenko OO, Hymanyk IV, Korol AR. Assessing the possibility of using portable and stationary non-mydriatic fundus cameras for diabetic retinopathy screening assisted by an artificial intelligence-based software platform in primary care. J Ophthalmol (Ukraine). 2024;6(521): 22-26. https://doi.org/10.31288/oftalmolzh202462226
Downloads
Published
How to Cite
Issue
Section
License
Copyright (c) 2025 Bezkorovaina I.M., Gontar A.R., Nakonechnyi D.O., Ivanchenko A.Iu.
This work is licensed under a Creative Commons Attribution 4.0 International License.
This work is licensed under a Creative Commons Attribution 4.0 International (CC BY 4.0) that allows users to read, download, copy, distribute, print, search, or link to the full texts of the articles, or use them for any other lawful purpose, without asking prior permission from the publisher or the author as long as they cite the source.
COPYRIGHT NOTICE
Authors who publish in this journal agree to the following terms:
- Authors hold copyright immediately after publication of their works and retain publishing rights without any restrictions.
- The copyright commencement date complies the publication date of the issue, where the article is included in.
DEPOSIT POLICY
- Authors are permitted and encouraged to post their work online (e.g., in institutional repositories or on their website) during the editorial process, as it can lead to productive exchanges, as well as earlier and greater citation of published work.
- Authors are able to enter into separate, additional contractual arrangements for the non-exclusive distribution of the journal's published version of the work with an acknowledgement of its initial publication in this journal.
- Post-print (post-refereeing manuscript version) and publisher's PDF-version self-archiving is allowed.
- Archiving the pre-print (pre-refereeing manuscript version) not allowed.
